Actually, it appears that what we hear is a combination of the in-room-measured steady-state response in the low frequencies, and the loudspeaker direct sound in the mid and high frequencies. On the other hand, the in-room-measured steady-state response is a combination of direct sound and various reflections across the spectrum, and depends mostly on loudspeaker directivity and room specifics.
This is easy to test - EQ two sufficiently different loudspeakers to the same target curve in-room and they will probably still sound very different (while now measuring a similar steady-state response).
This is why the ideal EQ strategy for loudspeakers is to EQ the low frequencies (usually under 300Hz) based on the in-room-measured steady-state response, and to EQ the rest based on anechoic data (which is not needed if the loudspeakers already measure well anechoically
).
That implies that the low-frequency EQ is room and setup dependent, but mid and high frequency EQ is derived from anechoic data and can therefore actually be universal for a specific loudspeaker model (assuming good unit-to-unit consistency - which is of course not always the case).
See also this thread by
@thewas for more thoughts from dr. Toole:
link.
One quote:
(
Source)